Jbase segmentation violation

[sivavi...@gmail.com]

Hi:

I like to know under what cirumstances the jbase segmentation
violation will occur apart from the below scenarios,

1. Signal failure during comitting
2. Overload or leakage of memory
3. dumping of huge data in an unassigned variable.

For me, the problem of jbase segmentation violation is occuring during
TRANSEND statement used in T24. It is not coming during all TRANSEND,
only while authorising specific contract(eg.SEC.TRADE). Also it is not
coming while authorising all contracts, during some contracts, this
error is coming. But the contract is similar to all the other
contracts. no difference.

Only thing, we can notice while this problem happening is "the desktop
hangs".

>From the tracers, we could able to find out that the system is hanging
due to jbase segmentation violation error from the com log.

Please let me know, what is the situation that could lead this problem
in jbase. Is there any jbase level tracers to log the error details
during segmentation error.

My jbase release - 4.1.5.8
Globus release - G13.2.07

ultimately, i need jbase level tracers to log the details of the
error.

Any one please help me.

Cheers
Siva.

[Jim Idle]

If you get segmentation violations, then it is a very severe jBASE bug.
Segmentation errors happened when the program is trying to access some
memory that it did not allocate, either because it has already free()d
it or because it has run off the end of a buffer, or the programmer just
lost a pointer somewhere.

You need to report this to TEMENOS with the highest priority, especially
if you can reproduce it.

Things that can happen that cause this, that you can fix though are when
a jBASE file gets very corrupt. Make sure you run jcheck on your hash
files. You MUST do this when there are no jBASE processes running on the
system, so it has to be offline. You can however run a fake jbackup,
which may crash on a corrupt file, and you can run jcheck when users are
on the system but if it reports a corrupt file, you need to check that
it really is corrupt by checking it offline or at least running jcheck a
number of times to see if the result is consistent. Search the group
archive for more information on this:
http://groups.google.com/group/jBASE/search?group=jBASE&q=jcheck

Also, there seems to have been a fair bit of work performed on the
transaction journaling in the release you mention, and so far my
experience with it is that it does not seem to be well tested. If this
is happening exclusively in transaction commits, then that may be your
problem.

Jim

[s7bd]

Not that your application should be doing this but you can readily
cause a segmentation violation by simply writing a record to a jBASE
distributed file configured to use a subroutine for it's hash and make
the key length > 1024. We ran across this in a legacy conversion
where some of the large hashed files had some remnants from corruption
in years past resulting in records with very large keys. When we
tried to import them into a jBASE distributed file, we got the
segfaults and hang. This problem will not occur with normal hashed
files, just distributed files. My personal opinion is that keys that
large, if not accepted by the database, should error gracefully and
not segfault but so far mine seems to be the minority opinion.
-BD

> Of sivavishn...@gmail.com

> Sent: Friday, May 11, 2007 6:26 AM
> To: jBASE

> Cc: sivavishn...@gmail.com

[Jim Idle]

This should definitely be reported as a bug. Just because you shouldn't
be doing something (old Tommy Cooper jokes not withstanding - a
reference only the Brits will get), doesn't mean the system should
crash, you should get a reasonable error message, and without an ON
ERROR clause, should drop in to the BASIC debugger (or exit on a live
system etc).

[Jim Myers]

Hi,

 

We just migrated from Universe on SCO-Unix to jBase on Linux Redhat.

 

Anytime We issue a server print command, we get a JBASE SEGMENTATION VIOLATION ERROR.

 

This has been the issue for the past month and a half.

 

How do i solve this problem?

 

JBASE VERSION PARTICULERS

 

System Information

==================

System : Linux linuxsys 2.4.21-4.ELsmp.#1 SMP Fri Oct 3 176

OS Release : Red Hat Enterprise Linux AS release 3 (Taroon)

UNIX User : jtest (uid 503, euid 503)

Tty name : /dev/pts/2

Time : Sat Dec 9 19:41:06 2006

Environment

===========

JBCPORTNO : Not Set

JBCRELEASEDIR : '/usr/jbc'

JBCGLOBALDIR : '/usr/jbc'

JBCDATADIR : '/usr/NDMIGRATION/migrate/bnk/bnk.run/jbase_data'

HOME : '/usr/NDMIGRATION/migrate/bnk/bnk.run'

JEDIFILEPATH : '/usr/NDMIGRATION/migrate/bnk/bnk.run'

JEDIFILENAME_MD : 'VOC'

JEDIFILENAME_SYSTEM : '/usr/jbc/src/SYSTEM'

RELEASE Information : Major 4.1 , Minor 5.10 , Patch 5521 (Change 41066)

Spooler dir (JBCSPOOLERDIR) : '/usr/NDMIGRATION/migrate/bnk/bnk.run/jspooler'

JBCEMULATE : 'prime'

Object path (JBCOBJECTLIST) : '/usr/NDMIGRATION/migrate/bnk/bnk.run/globuslib:/'

Compiler : gcc gcc (GCC) 3.2.3 20030502 (Red Hat Linux 3.2.3)

jBASE Compiler Run-time : '/usr/jbc/config/system.properties'

Program dir (JBCDEV_BIN) : '/usr/NDMIGRATION/migrate/bnk/bnk.run/bin'

Subroutine dir (JBCDEV_LIB) : '/usr/NDMIGRATION/migrate/bnk/bnk.run/lib'

Max open files : 1024

There are No warnings, jBASE seems to be loaded correctly

jsh jtest ~ -->

 

--
Jim Jerry Myers
Global Solutions Limited.
Cantoment
Tel: 0244722960
     0207394212

[raphael olapade]

Hi,

It has been observed that in Jbase4 there's segmentation issue, i.e.

We have filesize limitation in Jbase 4 as 2GB , files more than 2 Giga Bytes will give Jbase segmentation violation.

So you need to find out "What is the allowable filesize in Jbase 5 ?" or the jbase you are using.

Maybe what you are trying to print is more than 2 Giga Bytes"

Is it more than the existing 2GB ? Please find out.

thanks.


Sege

--
Olapade R.S
2, Akinfemi Salvage Str
Fagba Bus Stop, Iju RD, Lagos
Tel No:234-1-(0)8033087434/8051144422